Multimodal Image Registration Using GPU Parallel Computing Technology
Waterloo, Ontario, Canada
Master of Mathematics in Computational Mathematics Thesis,Waterloo, Ontario, Canada, 2010
@article{sun2010multimodal,
title={Multimodal Image Registration Using GPU Parallel Computing Technology},
author={Sun, X.},
year={2010}
}
This research project studies the parallel computing technique offered by the graphics processing unit (GPU), and uses it to accelerate the computation of image registration. Image registration is a process that aligns two images so that the point in one image corresponds to the same anatomical point in the other. It is a key part in the medical imaging analysis. Medical images are often taken at different time and places, resulting in varying frame of references for the same part of the human body in the images. As shown in Figure 1-1, a patient’s brain appears at the upper part in one image, whereas the same brain appears at the lower right part in the other image that is taken at a different time. To better detect the change of the patient’s brain over time, it is necessary to move one image to align with the other so that the differences between the two images can be easily identified.
March 5, 2011 by hgpu